conftrace_

2017 NeurIPS NeurIPS 2017

On the Consistency of Quick Shift

Abstract

Quick Shift is a popular mode-seeking and clustering algorithm. We present finite sample statistical consistency guarantees for Quick Shift on mode and cluster recovery under mild distributional assumptions. We then apply our results to construct a consistent modal regression algorithm.

Authors

Heinrich Jiang

Topics

Machine Learning > Core Methods > Clustering Machine Learning > Learning Types > Unsupervised Learning Machine Learning > Optimization & Theory > Theory Machine Learning > Optimization & Theory > Statistics

Keywords

statistical consistency cluster analysis nonparametric statistics kernel density estimation clustering algorithm modal regression quick shift

More from NeurIPS 2017

All NeurIPS 2017 papers →

Differentially Private Empirical Risk Minimization Revisited: Faster and More General

Unbounded cache model for online language modeling with open vocabulary

Shape and Material from Sound

Wasserstein Learning of Deep Generative Point Process Models

Machine Learning with Adversaries: Byzantine Tolerant Gradient Descent

Protein Interface Prediction using Graph Convolutional Networks